Biology subjects

Beck, T.

Publications and source records attributed to Beck, T..

At least 19 recordsLinked to original sources

bioRxiv2026

Relation Extraction for Diet, Non-Communicable Disease and Biomarker Associations (RECoDe): A CoDiet study

Diet plays a critical role in human health, with growing evidence linking dietary habits to disease outcomes. However, extracting structured dietary knowledge from biomedical literature remains challenging due to the lack of dedicated relation extraction datasets. To address this gap, we introduce RECoDe, a novel relation extraction (RE) dataset designed specifically for diet, disease, and related biomedical entities. RECoDe captures a diverse set of relation types, including a broad spectrum of positive association patterns and explicit negative examples, with over 5,000 human-annotated instances validated by up to five independent annotators. Furthermore, we benchmark various natural language processing (NLP) RE models, including BERT-based architectures and enhanced prompting techniques with locally deployed large language models (LLMs) to improve classification performance on underrepresented relation types. The best performing model gpt-oss-20B, a local LLM, achieved an F1-score of 64% for multi-class classification and 92% for binary classification using a hierarchical prompting strategy with a separate reflection step built in. To demonstrate the practical utility of RECoDe, we introduce the Contextual Co-occurrence Summarisation (Co-CoS) framework, which aggregates sentence-level relation extractions into document-level summaries and further integrates evidence across multiple documents. Co-CoS produces effect estimates consistent with established dietary knowledge, demonstrating its validity as a general framework for systematic evidence synthesis. AvailabilityThe code, models, and data will be made freely available upon acceptance.

bioinformatics↗

bioRxiv2025

Cells dynamically adapt their nuclear volumes and proliferation rates during single to multicellular transitions

Tumour development and progression are associated with biophysical alterations that manifest across multiple spatial scales, from the subcellular to multicellular tissue scale. While cells dynamically regulate their biophysical properties like volumes and mechanics in dependence of cell state and function, it is unclear how these properties are controlled in the dense multicellular environment of a developing tumour. Here, we quantified cell and nuclear volumes of single cancer cells, while they grew into multicellular tumour spheroids within well-defined, tuneable biohybrid polymer hydrogels. We quantitatively showed that the formation of multicellular structures is associated with marked reductions of cellular and nuclear volumes, cell cycle delays as well as cell mechanical alterations, and that these changes are coupled. Single-to-multicellular transitions coincided with a drastic decrease in median nuclear volumes by up to 60%, as well as overall cell volume decrease. Nuclear volume decrease could not be explained by compression due to confining microenvironments. Instead, cell cycle adaptions were one significant contributor, with smaller-sized G1 cells accumulating in growing clusters, an effect that was reversed by CDK1 inhibition. Another contributor was nuclear volume decrease in cells within clusters that was associated with higher mass density and stiffness and could be abrogated upon cell release from clusters. In turn, multicellular-to-single cell transitions that happened in cells that invaded from a tumour spheroid into the surrounding matrix, were accompanied by nuclear volume increases and cell softening. Taken together, our study provides insights into how cells dynamically adapt their cellular/nuclear volumes, cell cycle progression and mechanics in dependence of the multicellular state.

biophysics↗

bioRxiv2025

High-precision Biomedical Text Corpora for Multi-Entity Recognition: A CoDiet study

We present here five biomedical, multi-entity corpora that can be used as benchmarks for named-entity recognition (NER), targeted to literature on metabolic syndrome. The CoDiet-Gold corpus contains annotations for 500 full-text publications and 348,406 annotations. It is divided into CoDiet-Gold-public (450 documents) and CoDiet-Gold-private (50 documents). Each document was independently annotated by two human experts, with disagreements fully adjudicated by a third expert. The CoDiet-Electrum corpus (2,998,273 annotations) contains 4,423 publications that were annotated using case-insensitive matching of the surface forms with punctuation ignored, found in CoDiet-Gold-public. Finally, for the same 4,423 documents, two fully machine annotated corpora CoDiet-Bronze (2,938,738 annotations) and CoDiet-Silver (2,298,988 annotations), were created by utilising existing NER algorithms to annotate these. These corpora contain categories (organisms, disease, genes, proteins, metabolites) that add depth to existing corpora, as well as new categories that do not appear in other corpora (food, dietary methods, sample types, computational methods, study methodology, population characteristics, data types, and microbiome).

bioinformatics↗

bioRxiv2025

Microbial Named Entity Recognition and Normalisation for AI-assisted Literature Review and Meta-Analysis

MotivationManual curation of biomedical literature is slow and error-prone and while large language models (LLMs) trained on general texts have shown to be useful for text summarisation, these methods lack the domain-specific expertise required to perform this task accurately. Here we describe the creation of the first microbiome-specific text corpus, use this to train deep learning algorithms for named-entity recognition (NER) and normalisation (NEN), and demonstrate their use to meta-analyse microbiome literature. MethodsWe developed an automated pipeline to annotate all mentions of bacteria, archaea, and fungi in 1,410 full-text microbiome articles. We manually annotated (gold-standard) a separate test set of 288 documents. We trained different transformer-based language models for microbiome recognition and normalisation to taxonomic identifiers and evaluate their performance using the precision, recall, F1-score, and accuracy on the test set. The best models were used to automatically annotate all available Open Access, full-text microbiome articles (n=6,927) and identify taxa that are significantly overrepresented across 14 domains. ResultsThe training and validation set contained a total of 90,150 annotations (both long form and abbreviations). Using the gold-standard test set, with an inter-annotator agreement rate of 99.52% for NER and 88.31% for NEN, the trained models were evaluated and our fine-tuned BioBERT model achieved an F1-score of 96% for NER surpassing a rule- and dictionary-based annotation pipeline (94%). For NEN the accuracy obtained by the deep learning models greatly surpassed that of the pipeline (91% vs 69%). Evaluated across the entire available literature, our models annotate an entire full-text document in only 7 seconds. ConclusionOur algorithms have near perfect precision and greatly speed up the process of annotating microbes in full-text articles. We demonstrated the capabilities of these methods by analysing the entire available literature and describe the taxa associated with each of the domains in our meta-analysis, and exemplify how these methods can be integrated into literature review workflows improving both the speed and accuracy of results. AvailabilityAll codes and data for automatic annotation, model training, and generation of taxonomic trees visualising the data will be made available following peer review with instructions on how to deploy the model on new texts from https://github.com/omicsNLP/microbELP.

bioinformatics↗

bioRxiv2024

Optical quantification of molecular interaction strength in protein condensates

Biomolecular condensates have been identified as a ubiquitous means of intracellular organization, exhibiting very diverse material properties. However, techniques to characterize these material properties and their underlying molecular interactions are scarce. Here, we introduce two optical techniques - Brillouin microscopy and quantitative phase imaging (QPI) - to address this scarcity. We establish Brillouin shift and linewidth as measures for average molecular interaction and dissipation strength, respectively, and we used QPI to obtain the protein concentration within the condensates. We monitored the response of condensates formed by FUS and by the low-complexity domain of hnRNPA1 (A1-LCD) to altering temperature and ion concentration. Conditions favoring phase separation increased Brillouin shift, linewidth, and protein concentration. In comparison to solidification by chemical crosslinking, the ion-dependent aging of FUS condensates had a small effect on the molecular interaction strength inside. Finally, we investigated how sequence variations of A1-LCD, that change the driving force for phase separation, alter the physical properties of the respective condensates. Our results provide a new experimental perspective on the material properties of protein condensates. Robust and quantitative experimental approaches such as the presented ones will be crucial for understanding how the physical properties of biological condensates determine their function and dysfunction.

biophysics↗

bioRxiv2023

Estimation of the mass density of biological matter from refractive index measurements

The quantification of physical properties of biological matter gives rise to novel ways of understanding functional mechanisms by utilizing models that explicitly depend on physical observables. One of the basic biophysical properties is the mass density (MD), which determines the degree of crowdedness. It impacts the dynamics in sub-cellular compartments and further plays a major role in defining the opto-acoustical properties of cells and tissues. As such, the MD can be connected to the refractive index (RI) via the well known Lorentz-Lorenz relation, which takes into account the polarizability of matter. However, computing the MD based on RI measurements poses a challenge as it requires detailed knowledge of the biochemical composition of the sample. Here we propose a methodology on how to account for a priori and a posteriori assumptions about the biochemical composition of the sample as well as respective RI measurements. To that aim, we employ the Biot mixing rule of RIs alongside the assumption of volume additivity to find an approximate relation of MD and RI. We use Monte-Carlo simulations as well as Gaussian propagation of uncertainty to obtain approximate analytical solutions for the respective uncertainties of MD and RI. We validate this approach by applying it to a set of well characterized complex mixtures given by bovine milk and intralipid emulsion. Further, we employ it to estimate the mass density of trunk tissue of living zebrafish (Danio rerio) larvae. Our results enable quantifying changes of mass density estimates based on variations in the a priori assumptions. This illustrates the importance of implementing this methodology not only for MD estimations but for many other related biophysical problems, such as mechanical measurements using Brillouin microscopy and transient optical coherence elastography.

biophysics↗

bioRxiv2023

Identification of genetic drivers of plasma lipoproteins in the Diversity Outbred mouse population

Despite great progress in understanding lipoprotein physiology, there is still much to be learned about the genetic drivers of lipoprotein abundance, composition, and function. We used ion mobility spectrometry to survey 16 plasma lipoprotein subfractions in 500 Diversity Outbred (DO) mice maintained on a Western-style diet. We identified 21 quantitative trait loci (QTL) affecting lipoprotein abundance. To refine the QTL and link them to disease risk in humans, we asked if the human homologues of genes located at each QTL were associated with lipid traits in human genome-wide association studies (GWAS). Integration of mouse QTL with human GWAS yielded candidate gene drivers for 18 of the 21 QTL. This approach enabled us to nominate the gene encoding the neutral ceramidase, Asah2, as a novel candidate driver at a QTL on chromosome 19 for large HDL particles (HDL-2b). To experimentally validate Asah2, we surveyed lipoproteins in Asah2-/-mice. Compared to wild-type mice, female Asah2-/- mice showed an increase in several lipoproteins, including HDL. Our results provide insights into the genetic regulation of circulating lipoproteins, as well as mechanisms by which lipoprotein subfractions may affect cardiovascular disease risk in humans.

genetics↗

bioRxiv2023

Vocabulary Matters: An Annotation Pipeline and Two Deep Learning Algorithms for Enzyme Named Entity Recognition

Enzymes are indispensable substances in many biological processes. With biomedical literature growing exponentially, it becomes more difficult to review the literature effectively. Hence, text-mining techniques are needed to facilitate and speed up literature review. The aims of this study are to create a corpus with annotated enzymes to train and evaluate enzyme named-entity recognition (NER) models. A novel pipeline was built using a combination of dictionary matching and rulebased keyword searching to automatically annotate enzyme entities in over 4,800 biomedical full texts. Two Bidirectional Long Short-Term Memory (BiLSTM) networks using BioBERT and SciBERT as tokeniser and word embedding layers were trained on this corpus and evaluated on a manually annotated test set of 526 full-text publications. The dictionary- and rule-based annotation pipeline achieved an F1-score of 0.863 (precision 0.996, recall 0.762). The SciBERT-BiLSTM model (F1-score 0.965, precision 0.981, recall 0.954) largely out-performed the BioBERT-BiLSTM model (F1-score 0.955, precision 0.981, recall 0.937). This study contributed a novel dictionary- and rule-based automatic pipeline with almost perfect precision which runs in a matter of seconds on a standard laptop. Both deep learning (DL) models achieved state-of-the-art performance (F1>0.95) for enzyme NER, with the SciBERT-based model outperforming the BioBERT-based model in terms of recall, demonstrating the vocabulary used by models matters. The proposed pipeline with the DL models can facilitate more effective enzyme text-mining and information extraction research for literature review and are the first algorithms specifically for enzyme NER. AvailabilityAll codes are available for automatic annotation and model training (including data), with instructions on how to deploy the model on new text, from https://github.com/omicsNLP/enzymeNER.

bioinformatics↗

bioRxiv2023

Predominant myosin super-relaxed state in canine myocardium with naturally occurring dilated cardiomyopathy

Dilated cardiomyopathy (DCM) is a naturally occurring heart failure condition in humans and dogs, notably characterized by a reduced contractility and ejection fraction. As the identification of its underlying cellular and molecular mechanisms remain incomplete, the aim of the present study was to assess whether the molecular motor myosin and its known relaxed conformational states are altered in DCM. For that, we dissected and skinned thin cardiac strips from left ventricle obtained from six DCM Doberman Pinschers and six non-failing controls (NF). We then used a combination of Mant-ATP chase experiments and X-ray diffraction to assess both energetic and structural changes of myosin. Using the Mant-ATP chase protocol, we observed that in DCM dogs, the amount of myosin molecules in the ATP-conserving conformational state also known as super-relaxed (SRX), is significantly increased when compared with NF dogs. This alteration can be rescued by applying EMD-57033, a small molecule activating myosin. Conversely, with X-ray diffraction, we found that in DCM dogs, there is a higher proportion of myosin heads in the vicinity of actin when compared with NF dogs (1,0 to 1,1 intensity ratio). Hence, we observed an uncoupling between energetic (Mant-ATP chase) and structural (X-ray diffraction) data. Taken together, these results may indicate that in the heart of Doberman Pinschers with DCM, myosin molecules are potentially stuck in a non-sequestered but ATP-conserving SRX state, that can be counterbalanced by EMD-57033 demonstrating the potential for a myosin-centered pharmacological treatment of DCM. New & noteworthyThe key finding of the present study is that, in left ventricles of dogs with a naturally occurring dilated cardiomyopathy, relaxed myosin molecules favor a non-sequestered super-relaxed state potentially impairing sarcomeric contractility. This alteration is rescuable by applying a small molecule activating myosin known as EMD-57033.

pathology↗

bioRxiv2022

Small leucine-rich proteoglycans inhibit CNS regeneration by modifying the structural and mechanical properties of the lesion environment

Extracellular matrix (ECM) deposition after central nervous system (CNS) injury leads to inhibitory scarring in mammals, whereas it facilitates axon regeneration in the zebrafish. However, the molecular basis of these different fates is not understood. Here, we identify small leucine-rich proteoglycans (SLRPs) as a causal factor in regeneration failure. We demonstrate that the SLRPs Chondroadherin, Fibromodulin, Lumican, and Prolargin are enriched in human, but not zebrafish, CNS lesions. Targeting SLRPs to the zebrafish injury ECM inhibits axon regeneration and functional recovery. Mechanistically, we find that SLRPs confer structural and mechanical properties to the lesion environment that are adverse to axon growth. Our study reveals SLRPs as previously unknown inhibitory ECM factors in the human CNS that impair axon regeneration by modifying tissue mechanics and structure. ONE SENTENCE SUMMARYComposition, structural organization, and mechanical properties of the injury ECM direct central nervous system regeneration.

neuroscience↗

bioRxiv2022

A method for stabilising the XX karyotype in female mESC cultures

Female mouse embryonic stem cells (mESCs) present differently to male mESCs in several fundamental ways, however complications with their in vitro culture have resulted in an underrepresentation of female mESCs in the literature. Recent studies show that the second X chromosome in female, and more specifically the transcriptional activity from both of these chromosomes due to absent X chromosome inactivation, sets female mESCs apart from males. To avoid this undesirable state, female mESCs in culture preferentially adopt an XO karyotype, with this adaption leading to loss of their unique properties in favour of a state near indistinguishable from male. If female pluripotency is to be studied effectively in this system, it is critical that high quality cultures of XX mESCs are available. Here we report a method for better maintaining XX female mESCs in culture, which also stabilises the male karyotype, and makes study of female specific pluripotency more feasible.

cell biology↗

bioRxiv2022

In vivo assessment of mechanical properties during axolotl development and regeneration using confocal Brillouin microscopy

In processes such as development and regeneration, where large cellular and tissue rearrangements occur, cell fate and behavior are strongly influenced by tissue mechanics. While most well-established tools probing mechanical properties require an invasive sample preparation, confocal Brillouin microscopy captures mechanical parameters optically with high resolution in a contact-free and label-free fashion. In this work, we took advantage of this tool and the transparency of the highly regenerative axolotl to probe its mechanical properties in vivo for the first time. We mapped the Brillouin frequency shift with high resolution in developing limbs and regenerating digits, the most studied structures in the axolotl. We detected a gradual increase in the cartilage Brillouin frequency shift, suggesting decreasing tissue compressibility during both development and regeneration. Moreover, we were able to correlate such increase with the regeneration stage, which was undetected with fluorescence microscopy imaging. The present work evidences the potential of Brillouin microscopy to unravel the mechanical changes occurring in vivo in axolotls, setting the basis to apply this technique in the growing field of epimorphic regeneration.

developmental biology↗

bioRxiv2022

MetaboListem and TABoLiSTM: Two Deep Learning Algorithms for Metabolite Named Entity Recognition

Reviewing the metabolomics literature is becoming increasingly difficult because of the rapid expansion of relevant journal literature. Text-mining technologies are therefore needed to facilitate more efficient literature review. Here we contribute a standardised corpus of full-text publications from metabolomics studies and describe the development of two new metabolite named entity recognition (NER) methods. We introduce two deep learning methods for metabolite NER based on Bidirectional Long Short-Term Memory (BiLSTM) networks incorporating different transfer learning techniques. Our first model (MetaboListem) follows prior methodology using GloVe word embeddings. Our second model exploits BERT and BioBERT for embedding and is named TABoLiSTM (Transformer-Affixed BiLSTM). The methods are trained on a novel corpus annotated using rule-based methods, and evaluated on manually annotated metabolomics articles. MetaboListem (F1 score 0.890, precision 0.892, recall 0.888) and TABoLiSTM (BioBERT version: F1 score 0.909, precision 0.926, recall 0.893) have achieved state-of-the-art performance on metabolite NER. A corpus with >1,200 full-text Open Access metabolomics publications and >116,000 annotated metabolites was created. This work demonstrates that deep learning algorithms are capable of identifying metabolite names accurately and efficiently in text. The proposed corpus and NER algorithms can be used for metabolomics text-mining tasks such as information retrieval, document classification and literature-based discovery. AvailabilityThe corpus and NER algorithms are freely available with detailed instructions from Github at https://github.com/omicsNLP/MetaboliteNER.

bioinformatics↗

bioRxiv2021

Maternal SMCHD1 controls both imprinted Xist expression and imprinted X chromosome inactivation

Embryonic development is dependent on the maternal supply of proteins through the oocyte, including factors setting up the adequate epigenetic patterning of the zygotic genome. We previously reported that one such factor is the epigenetic repressor SMCHD1, whose maternal supply controls autosomal imprinted expression in mouse preimplantation embryos and mid-gestation placenta. In mouse preimplantation embryos, X chromosome inactivation is also an imprinted process. Combining genomics and imaging, we show that maternal SMCHD1 is required not only for the imprinted expression of Xist in preimplantation embryos, but also for the efficient silencing of the inactive X in both the preimplantation embryo and mid-gestation placenta. These results expand the role of SMCHD1 in enforcing the silencing of Polycomb targets. The inability of zygotic SMCHD1 to fully restore imprinted X inactivation further points to maternal SMCHD1s role in setting up the appropriate chromatin environment during preimplantation development, a critical window of epigenetic remodelling.

genomics↗

bioRxiv2021

Mapping tumor spheroid mechanics in dependence of 3D microenvironment stiffness and degradability by Brillouin microscopy

Altered biophysical properties of cancer cells and of their microenvironment contribute to cancer progression. While the relationship between microenvironmental stiffness and cancer cell mechanical properties and responses has been previously studied using two-dimensional (2D) systems, much less is known about it in a physiologically more relevant 3D context and in particular for multicellular systems. To investigate the influence of microenvironment stiffness on tumor spheroid mechanics, we first generated MCF-7 tumor spheroids within matrix metalloproteinase (MMP)-degradable 3D polyethylene glycol (PEG)-heparin hydrogels, where spheroids showed reduced growth in stiffer hydrogels. We then quantitatively mapped the mechanical properties of tumor spheroids in situ using Brillouin microscopy. Maps acquired for tumor spheroids grown within stiff hydrogels showed elevated Brillouin frequency shifts (hence increased longitudinal elastic moduli) with increasing hydrogel stiffness. Maps furthermore revealed spatial variations of the mechanical properties across the spheroids cross-sections. When hydrogel degradability was blocked, comparable Brillouin frequency shifts of the MCF-7 spheroids were found in both compliant and stiff hydrogels, along with similar levels of growth-induced compressive stress. Under low compressive stress, single cells or free multicellular aggregates showed consistently lower Brillouin frequency shifts compared to spheroids growing within hydrogels. Thus, the spheroids mechanical properties were modulated by matrix stiffness and degradability as well as multicellularity, and also to the associated level of compressive stress felt by tumor spheroids. Spheroids generated from a panel of invasive breast, prostate and pancreatic cancer cell lines within degradable stiff hydrogels, showed higher Brillouin frequency shifts and less cell invasion compared to those in compliant hydrogels. Taken together, our findings contribute to a better understanding of the interplay between cancer cells and microenvironment mechanics and degradability, which is relevant to better understand cancer progression.

biophysics↗

bioRxiv2021

Maternal SMCHD1 regulates Hox gene expression andpatterning in the mouse embryo

Parents transmit genetic and epigenetic information to their offspring. Maternal effect genes regulate the offspring epigenome to ensure normal development. Here we report that the epigenetic regulator SMCHD1 has a maternal effect on Hox gene expression and skeletal patterning. Maternal SMCHD1, present in the oocyte and preimplantation embryo, prevents precocious activation of Hox genes postimplantation. Without maternal SMCHD1, highly penetrant posterior homeotic transformations occur in the embryo. Hox genes are decorated with Polycomb marks H2AK119ub and H3K27me3 from the oocyte throughout early embryonic development; however, loss of maternal SMCHD1 does not alter these marks. Therefore, we propose maternal SMCHD1 acts downstream of Polycomb marks to establish a chromatin state necessary for persistent epigenetic silencing and appropriate Hox gene expression later in the developing embryo. This is a striking role for maternal SMCHD1 in long-lived epigenetic effects impacting offspring phenotype.

developmental biology↗

bioRxiv2021

SMCHD1 has separable roles in chromatin architecture and gene silencing that could be targeted in disease

The interplay between 3D chromatin architecture and gene silencing is incompletely understood. Here, we report a novel point mutation in the non-canonical SMC protein SMCHD1 that enhances its silencing capacity at endogenous developmental targets and at the facioscapulohumeral muscular dystrophy associated macro-array, D4Z4. Heightened SMCHD1 silencing perturbs developmental Hox gene activation, causing a homeotic transformation in mice. Paradoxically, the mutant SMCHD1 appears to enhance insulation against another epigenetic regulator complex, PRC2, while depleting long range chromatin interactions akin to what is observed in the absence of SMCHD1. These data suggest that SMCHD1s role in long range chromatin interactions is not directly linked to gene silencing or insulating the chromatin, refining the model for how the different levels of SMCHD1-mediated chromatin regulation interact to bring about gene silencing in normal development and disease.

molecular biology↗

bioRxiv2021

Auto-CORPus: Automated and Consistent Outputs from Research Publications

To analyse large corpora using machine learning and other Natural Language Processing (NLP) algorithms, the corpora need to be standardised. The BioC format is a community-driven simple data structure for sharing text and annotations, however there is limited access to biomedical literature in BioC format and a lack of bioinformatics tools to convert online publication HTML formats to BioC. We present Auto-CORPus (Automated pipeline for Consistent Outputs from Research Publications), a novel NLP tool for the standardisation and conversion of publication HTML and table image files to three convenient machine-interpretable outputs to support biomedical text analytics. Firstly, Auto-CORPus can be configured to convert HTML from various publication sources to BioC. To standardise the description of heterogenous publication sections, the Information Artifact Ontology is used to annotate each section within the BioC output. Secondly, Auto-CORPus transforms publication tables to a JSON format to store, exchange and annotate table data between text analytics systems. The BioC specification does not include a data structure for representing publication table data, so we present a JSON format for sharing table content and metadata. Inline tables within full-text HTML files and linked tables within separate HTML files are processed and converted to machine-interpretable table JSON format. Finally, Auto-CORPus extracts abbreviations declared within publication text and provides an abbreviations JSON output that relates an abbreviation with the full definition. This abbreviation collection supports text mining tasks such as named entity recognition by including abbreviations unique to individual publications that are not contained within standard bio-ontologies and dictionaries. AvailabilityThe Auto-CORPus package is freely available with detailed instructions from Github at https://github.com/omicsNLP/Auto-CORPus/.

bioinformatics↗